Control Panel

Place a Call

Phone Number (E.164 format)

Language

Last Call

—

initiated

Post-Call Analysis

Duration

—

User Turns

—

Bot Turns

—

Barge-ins

—

—✓ —✗

Conversation Timeline

No analysis loaded

✓ Accepted

✗ Rejected (too short / low-confidence)

Recent Calls

No calls yet

LLM Chat Test

Type a message to test the LLM using current prompt & settings.

Uses current Settings (temperature, max tokens) and Prompts tab system prompt.

🇺🇸 English Prompt

Saved changes take effect on the next call. Use {date} for today's date.

System Prompt

0 chars

Initial Message (bot's first line)

🇯🇵 Japanese Prompt

Saved changes take effect on the next call. Use {date} for today's date.

System Prompt

0 chars

Initial Message (bot's first line)

📜 Version History

Model & Call Settings

Temperature 0.67

0 — precise2 — creative

Max Tokens

200–500 recommended for voice calls.

TTS Speed 0.92

0.5 — slow2.0 — fast

Default Inbound Language

🇺🇸 English 🇯🇵 Japanese

Used when the INBOUND_LG env var is not set.

Prompt Evaluation

Auto-tests the active prompt — the LLM acts as judge and scores each response

8 default scenarios • click ✨ to generate prompt-specific tests

—

Pass Rate

—

Avg Score

⚠ Recommendations

📝 Custom Test Scenarios

Loading…

📈 Eval History

Score Trend — avg per run, oldest → newest

⚖️ A/B Prompt Comparison

Prompt A — Saved Version

Prompt B — Current Active

Saved prompt (auto-loaded)

Runs all active scenarios against both versions sequentially

🎭 Conversation Simulator

Define a conversation script and auto-run it against the active bot prompt

Script — one user turn per row

Turn 1